Originally posted by Squier Hi Dave - I was using MF only , not using AF to confirm focus.
I focused just once for the first shot,. I then didnt touch it again
Squier,
I happened to spend the better part of today calibrating a split-image focus-screen using a ruler. I took about 100 test shots. (Note: Body was K-m. Used a tripod and 2s timer.)
Here are some things I learned:
- Repetition. I ended up repeating my focus test at least 3 or 4 times per condition. Even with a split-image screen and a f/1.4 lens and a 1.2x eyepiece magnifier I occasionally had 3 different results from 3 different focusing attempts (sometimes I would have FF, BF, FF). That tells me that even with those precise focusing aids and narrow viewfinder DOF, my eye was not good enough to focus 100%. (Also see below about focusing direction.) Therefore, with the stock-screen and/or less bright lenses the accuracy will be less.
- The test should be done at the minimum focus distance of the lens. The reasons are (i) the farther something is away then the greater DOF at a given distance and (ii) it's hard to see the ruler when it is far away. Even at 1:7 magnification (typical minimum focus distance of 50/1.4mm lens), the 1cm increments on a ruler are pretty small in the viewfinder and I had trouble getting repeatable results. With DA35/2.8 macro lens, when I tried to test it at a similar magnification as the 50/1.4 I again had variable results. But at 1:1 it was much easier because the DOF was narrower and the ruler-lines appeared larger.
- Absolute BF/FF measurements with a ruler have meaning only at a given magnification. For example, at 1:7 I might see a FF/BF variation of up to 10mm which seems a lot. But at 1:1, I would see a variation of <1mm.
- FF/BF can vary by lens (and by zoom position). I used 5 different prime lenses (DA35/2.8, A50/1.4, VL58/1.4, K135/2.5, K24/2.8) and one zoom at 2 lengths (DAL18-55@35 and @55). For example, the DA35 consistently BF'd versus the A50/1.4. Another example, the DAL @55 consistently BF'd versus @35. (I don't know what to make of this. I knew that AF lenses could have variations in FF/BF, but I thought with MF that WYSIWYG. Maybe because the greater DOF's (f/2.8 vs. f/1.4 and f/5.6 vs. f/4.5), I couldn't tell? )
- Focusing direction makes a difference! I found that if I started focusing starting from from minimum distance direction, then I would be more likely to front-focus. If I focused starting from infinity direction, then I would be more likely to back-focus. The greater the DOF in the viewfinder, the worse this should be. Try focusing test 3x from both direction, see what happens. I didn't notice this until close to the end of my test sequence, so it's annoying I can't eliminate that variable because each focus test was uncontrolled in direction. However, I would usually go back and forth (like an AF lens, whirr, whirr, whirr)...but it's possible I didn't always which would account for the outliers in my data.
In conclusion:
1) There are a lot of ways to skew a focusing-test.
2) Even on a calibrated system, a very small amount of BF/FF is difficult to avoid. With 50/1.4 lens at f/1.4 and 1:7x, about 5~7mm is something I could live with because under the most controlled conditions I can't do better than that.